Help wanted

Run a model for the Index

There are too many models and effort levels for us to run alone. If we gave you an invite code, you can run any of these with your own API keys or subscriptions. Your results go on the Verified board, run by the same locked-down runner we use.

Fast results1 / 128model × effort pairs, 50 models
Claimed now2for 24 h each
Thorough0 / 1the top 5 on Fast

How to help

  1. Save your invite code (once, needs Node 20+). It asks whether to show your name next to your results.
    npx cube-arena-runner login AB12-CD34
  2. Pick a row below and run its command. It claims that model and effort for 24 hours, plays the Fast suite (about 45 minutes), and posts the result. Claude models run through Claude Code, GPT models through Codex, and anything else through OpenRouter (OPENROUTER_API_KEY) or another OpenAI-compatible API (--via api).
  3. That's it. The result shows up on the Index. If two people run the same thing, the Index shows the average. The runner also uploads a text log of each session (no pictures) so we can check it. If you show your name, you go on the contributors board, with your X handle if you add one (npx cube-arena-runner profile --x yourhandle).

No invite? You can still play any suite on the Community board: see Connect. Models come from Artificial Analysis's current list (as of 2026-10-11), kept to those that can read an image and call tools.

Fast suite

Every current model that can see the cube, at every effort it offers.

ModelEffortStatusReach it viaRun it
Claude Haiku 5.5 · AnthropiclowNeededClaude Code, OpenRouter
Claude Haiku 5.5 · AnthropicmediumNeededClaude Code, OpenRouter
Claude Haiku 5.5 · AnthropichighNeededClaude Code, OpenRouter
Claude Haiku 5.5 · AnthropicxhighNeededClaude Code, OpenRouter
Claude Haiku 5.5 · AnthropicmaxNeededClaude Code, OpenRouter
Mistral Large 4 Preview · MistralhighNeededOpenRouter
GPT-6.1-Sol · OpenAIlowNeededCodex, OpenRouter
GPT-6.1-Sol · OpenAImediumNeededCodex, OpenRouter
GPT-6.1-Sol · OpenAIhighNeededCodex, OpenRouter
GPT-6.1-Sol · OpenAIxhighNeededCodex, OpenRouter
GPT-6.1-Sol · OpenAImaxNeededCodex, OpenRouter
Claude Sonnet 5.5 · AnthropiclowNeededClaude Code, OpenRouter
Claude Sonnet 5.5 · AnthropicmediumNeededClaude Code, OpenRouter
Claude Sonnet 5.5 · AnthropichighNeededClaude Code, OpenRouter
Claude Sonnet 5.5 · AnthropicxhighNeededClaude Code, OpenRouter
Claude Opus 5.5 · AnthropiclowNeededClaude Code, OpenRouter
Claude Opus 5.5 · AnthropicmediumNeededClaude Code, OpenRouter
Claude Opus 5.5 · AnthropichighNeededClaude Code, OpenRouter
Claude Opus 5.5 · AnthropicxhighNeededClaude Code, OpenRouter
GPT-6-Luna · OpenAIlowNeededCodex, OpenRouter
GPT-6-Luna · OpenAImediumNeededCodex, OpenRouter
GPT-6-Luna · OpenAIhighNeededCodex, OpenRouter
GPT-6-Luna · OpenAIxhighNeededCodex, OpenRouter
GPT-6-Sol · OpenAIlowNeededCodex, OpenRouter
GPT-6-Sol · OpenAImediumNeededCodex, OpenRouter
GPT-6-Sol · OpenAIhighNeededCodex, OpenRouter
GPT-6-Sol · OpenAIxhighNeededCodex, OpenRouter
GPT-6-Sol · OpenAImaxNeededCodex, OpenRouter
Grok 4.7 · SpaceXAIlowNeededOpenRouter
Grok 4.7 · SpaceXAImediumNeededOpenRouter
Grok 4.7 · SpaceXAIhighNeededOpenRouter
Grok 4.7 · SpaceXAIxhighNeededOpenRouter
MiMo-V2.6-Flash · Xiaomi–NeededOpenRouter
MiMo-V2.6-Pro · Xiaomi–NeededOpenRouter
Step 5 Preview · StepFunlowNeededOpenRouter
Step 5 Preview · StepFunmediumNeededOpenRouter
Step 5 Preview · StepFunhighNeededOpenRouter
DeepSeek V4.1 Flash · DeepSeeklowNeededOpenRouter
DeepSeek V4.1 Flash · DeepSeekhighNeededOpenRouter
DeepSeek V4.1 Flash · DeepSeekmaxNeededOpenRouter
Ling-3.0-flash-VL · InclusionAI–NeededOpenRouter
GPT-6-Astra · OpenAIlowNeededCodex, OpenRouter
GPT-6-Astra · OpenAImediumNeededCodex, OpenRouter
GPT-6-Astra · OpenAIhighNeededCodex, OpenRouter
GPT-6-Astra · OpenAIxhighNeededCodex, OpenRouter
GPT-6-Astra · OpenAImaxNeededCodex, OpenRouter
Gemini 3.8 Flash · GooglelowNeededOpenRouter
Gemini 3.8 Flash · GooglemediumNeededOpenRouter
Gemini 3.8 Flash · GooglehighNeededOpenRouter
Muse Spark 1.3 · MetalowNeededOpenRouter
Muse Spark 1.3 · MetamediumNeededOpenRouter
Muse Spark 1.3 · MetahighNeededOpenRouter
Muse Spark 1.3 · MetaxhighNeededOpenRouter
Muse Spark 1.3 · MetamaxNeededOpenRouter
Qwen3.8 Max (0902) · AlibabalowNeededOpenRouter
Qwen3.8 Max (0902) · AlibabamediumNeededOpenRouter
Qwen3.8 Max (0902) · AlibabahighNeededOpenRouter
Qwen3.8 Max (0902) · AlibabaxhighNeededOpenRouter
Claude Fable 5.1 · AnthropiclowNeededClaude Code, OpenRouter
Claude Fable 5.1 · AnthropicmediumNeededClaude Code, OpenRouter
Claude Fable 5.1 · AnthropichighNeededClaude Code, OpenRouter
Claude Fable 5.1 · AnthropicxhighNeededClaude Code, OpenRouter
Claude Fable 5.1 · AnthropicmaxNeededClaude Code, OpenRouter
GLM-5.3-Flash · Z AIlowNeededOpenRouter
GLM-5.3-Flash · Z AIhighNeededOpenRouter
GLM-5.3-Flash · Z AImaxNeededOpenRouter
DeepSeek V4 Flash Vision · DeepSeeklowNeededOpenRouter
DeepSeek V4 Flash Vision · DeepSeekhighNeededOpenRouter
DeepSeek V4 Flash Vision · DeepSeekmaxNeededOpenRouter
Qwen3.8 27B · AlibabalowNeededOpenRouter
Qwen3.8 27B · AlibabamediumNeededOpenRouter
Qwen3.8 27B · AlibabaxhighNeededOpenRouter
Muse Glimmer · MetalowNeededOpenRouter
Muse Glimmer · MetamediumNeededOpenRouter
Muse Glimmer · MetahighNeededOpenRouter
Muse Glimmer · MetaxhighNeededOpenRouter
Inkling Small · Thinking MachineslowNeededOpenRouter
Inkling Small · Thinking MachinesmediumNeededOpenRouter
Inkling Small · Thinking MachineshighNeededOpenRouter
Inkling Small · Thinking MachinesmaxNeededOpenRouter
Gemini 3.5 Flash-Lite · GooglelowNeededOpenRouter
Gemini 3.5 Flash-Lite · GooglemediumNeededOpenRouter
Gemini 3.5 Flash-Lite · GooglehighNeededOpenRouter
Kimi K3 · KimilowNeededOpenRouter
Kimi K3 · KimihighNeededOpenRouter
Kimi K3 · KimimaxNeededOpenRouter
Inkling · Thinking MachineslowNeededOpenRouter
Inkling · Thinking MachinesmediumNeededOpenRouter
Inkling · Thinking MachineshighNeededOpenRouter
Inkling · Thinking MachinesmaxNeededOpenRouter
GPT-5.6-Terra · OpenAIlowNeededCodex, OpenRouter
GPT-5.6-Terra · OpenAImediumNeededCodex, OpenRouter
GPT-5.6-Terra · OpenAIhighNeededCodex, OpenRouter
GPT-5.6-Terra · OpenAIxhighNeededCodex, OpenRouter
GPT-5.6-Terra · OpenAImaxNeededCodex, OpenRouter
Kimi K2.7 Code · Kimi–NeededOpenRouter
MiniMax-M3 · MiniMax–NeededOpenRouter
Qwen3.7 Plus · Alibaba–NeededOpenRouter
Command A+ · Cohere–NeededOpenRouter
Mistral Medium 3.5 · MistralhighNeededOpenRouter
GPT-5.5 Pro · OpenAImediumNeededOpenRouter
GPT-5.5 Pro · OpenAIhighNeededOpenRouter
GPT-5.5 Pro · OpenAIxhighNeededOpenRouter
Qwen3.6 35B A3B · Alibaba–NeededOpenRouter
Gemma 4 26B A4B · Google–NeededOpenRouter
Gemma 4 31B · Google–NeededOpenRouter
Mistral Small 4 · MistralhighNeededOpenRouter
Qwen3.5 9B · Alibaba–NeededOpenRouter
Qwen3.5 122B A10B · Alibaba–NeededOpenRouter
Gemini 3.1 Pro Preview · GooglelowNeededOpenRouter
Gemini 3.1 Pro Preview · GooglemediumNeededOpenRouter
Gemini 3.1 Pro Preview · GooglehighNeededOpenRouter
Qwen3.5 397B A17B · Alibaba–NeededOpenRouter
GPT-5.3 Codex · OpenAIlowNeededOpenRouter
GPT-5.3 Codex · OpenAImediumNeededOpenRouter
GPT-5.3 Codex · OpenAIhighNeededOpenRouter
GPT-5.3 Codex · OpenAIxhighNeededOpenRouter
Ministral 3 14B · Mistral–NeededOpenRouter
Ministral 3 3B · Mistral–NeededOpenRouter
Ministral 3 8B · Mistral–NeededOpenRouter
Nova 2.0 Lite · Amazon–NeededOpenRouter
Nova Premier · Amazon–NeededOpenRouter
o3 · OpenAI–NeededOpenRouter
Llama 4 Maverick · Meta–NeededOpenRouter
Llama 4 Scout · Meta–NeededOpenRouter

Thorough suite

Up to about 8 hours, so only the 5 best models on Fast, at their top effort. This list fills in as Fast results arrive.

ModelEffortStatusReach it viaRun it
GPT-6-Luna · OpenAImaxNeededCodex, OpenRouter